Doom developer John Carmack thinks artificial general intelligence is doable by 2030
- Reference: 1695712506
- News link: https://www.theregister.co.uk/2023/09/26/john_carmack_agi/
- Source link:
Carmack shared his view at an event for the [1]announcement [Video] that his AGI startup Keen has hired Richard Sutton, chief scientific advisor at the Alberta Machine Intelligence Institute. Carmack described Sutton as "the godfather of reinforcement learning" and said that makes him ideally positioned to lead work on an AGI – an artificial intelligence that approximates human behavior.
Sutton told the event he does not think coding an AGI is infeasible with current techniques, adding that he sees a target of 2030 as possible for a "prototype AI to show signs of life."
[2]
During the event, Sutton and Carmack named a handful of Keen staff who will work on the project. Both acknowledged that Keen's team is tiny – as is the $20 million of funding it has secured, compared to the billions of dollars and hundreds of workers at larger AI outfits.
[3]
[4]
But Sutton and the Keen team are sufficiently audacious they believe they can make a contribution.
Carmack weighed in with his opinion that while "Nobody has line of sight on a solution to this today, we feel there is not that much left to do." Those remaining problems, he said, don't require entirely new thinking or architectures.
[5]
"There are fundamental research questions that need to be answered, and we have internal projects and angles of attack," he suggested, adding that he expects that future textbooks on how to build AGIs will include a chapter on the work Keen and others will do in coming years.
And he believes that effort will soon produce something that looks like success.
"We are six, seven, eight years out from something really big and important being publicly visible."
[6]Doom : The FPS that wowed players, gummed up servers, and enraged admins
[7]Doom and Super Mario could be a lot tougher now AI is building levels
[8]OpenAI challenges you to beat 1990s classic Sonic the Hedgehog using machine learning
[9]Unreal: Epic’s would-be Doom... er... Quake killer
In a subsequent Q&A session Carmack mused that the present state of AI may only be possible due to his work on seminal game Quake – which followed Doom, which followed Wolfenstein 3D – as it sparked demand for GPUs. He credited Nvidia's Jensen Huang with recognizing the GPU's potential to handle other computing chores, thereby sparking the current AI boom, but still took pride in his contribution.
Carmack also revealed that OpenAI CEO Sam Altman once tried to hire him, despite his complete lack of AI experience at the time. While he turned down the offer, Carmack became curious about AI. Monday's announcement was one result.
[10]
He's since boned up on the subject to the extent that he opined "Concerns that the monolithic nature of the models [used at the Alberta Machine Intelligence Institute] may not capture some of the human brain's work as a much more distributed semi-consensus-finding system.'
In computer science, he said, there's a tendency to draw "neat boxes." His opinion is that our sole proven model for how consciousness works – the human brain – shows that intelligence is more complex.
"The important thing is about how an AI can digest its experience into its view of the world and how it predicts things going forward and how it needs to have motivations both internally and externally imposed."
Those issues, he said, are not addressed by large language models, which don't address how our brains work. Yet every lab in the world, he said, is currently "throwing resources" at such models.
"Being a bit contrarian, both of us, I think is a positive thing," he added, with reference to Keen pursuing AGI. Carmack also observed that Keen has no deadline or pressure to deliver a product.
Phew. ®
Get our [11]Tech Resources
[1] https://www.youtube.com/watch?v=aM7F5kuMjRA
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZRKrwwmalkVaCnux@IU-lgAAAAc&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZRKrwwmalkVaCnux@IU-lgAAAAc&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZRKrwwmalkVaCnux@IU-lgAAAAc&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZRKrwwmalkVaCnux@IU-lgAAAAc&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2018/12/10/doom_turns_25/
[7] https://www.theregister.com/2018/05/01/ai_can_generate_new_levels_for_classic_video_games_like_super_mario_and_doom/
[8] https://www.theregister.com/2018/04/05/openai_challenge_developers_to_play_sonic_the_hedgehog_with_ai/
[9] https://www.theregister.com/2013/07/16/antique_code_show_unreal/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZRKrwwmalkVaCnux@IU-lgAAAAc&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://whitepapers.theregister.com/
There are two measure/targets for achieving AGI - a shorter term one to convince investors of the achievement, and the longer term one to actually achieve it.
Convincing investors is easy though. You just need a convincing demo...I could do that cheap if you like. I'll need a midget, a large box, two laptops, Discord and an internet connection.
I agree with Carmack here. It's doable.
LLMs are a product of limited hardware resources...it's the max that is possible with the technology available. What we have now does not represent the cutting edge of theoretical AI...what we have now is what is possible with a shit load of GPUs for training and the available resources on an average desktop. The training is limited to the sheer number of GPUs you can afford to buy and house in a single location...the usage of LLMs is limited to what is available on commodity servers...the existing solutions reek of compromise and trade offs...
All Carmack needs to do is figure out a way to train more complicated models with fewer resources, and produce models that don't require as much VRAM.
These are two problems he is very capable of solving as he has a long history of getting a lot out of not much. I think this is what Carmack can see...in the same way he saw smooth graphics and 3D rendering on machines in the late 80s / early 90s when nobody else could.
Whatever he does here, will be interesting and potentially ground breaking. The investment so far seems tiny, but given his background...it's actually massive.
Not bigger... not monolithic.
That sounds dangerously like "if we just make it bigger, it will generalise". I don't think that's true, and I don't think that's what Carmack is doing.
Meanwhile, a lot of the AI hype companies are focussing purely on making it bigger - because that's a thing they can explain to investors, because it's a convenient moat around their business, and because large models do indeed produce "finer grained" output which looks like progress. I think they are making better tools, but that's not the same as AGI.
No. The problem with LLMs is that they are language models, and not knowledge models. Doesn't matter how big you make them.
AGI says no
Imagine if they develop AGI and it just does nothing but throwing a tantrum at every occasion or gets itself busy watching Tik Toks.
They will also have to develop a way to administer it with virtual drugs, so it can keep being focused, "happy" and less shy.
I think therapists have a bright future.
At phone repair shop: "My phone assistant gave me wrong directions and it now keeps belittling me. I have really low self esteem now. Is it true that my face looks like Picasso's botched work?"
Repair person: "Don't worry, your face is just fine! This looks like a job for our phone therapist. I can schedule an appointment for your phone on Tuesday, does that work for you?"
Client: "Of course! Thank you so much!"
Repair person: "Okay, so I wrote you down the details. Don't let your phone assistant know. See you Next Tuesday!"
Re: AGI says no
doctor, my cat-master belittles me, can you book me a session too? Sure, just speak to my assistant!
consciousness v. intelligence
wonder if one is prerequisite for the other and which comes first. Both terms seem to (to me) used somewhat 'nonchalantly' in the context of AI.
All of this has happened before...
2030: All of humanity hails the first sentient AI;
2039: Sentient robots are used as slaves everywhere;
2040: Cylons rebel and attack their masters;
2041: Mass exodus of the seven continents?
Own goal
Notes found in the rubble of civilisation by little green archaeologists, "I am the last human alive. As I write this, the killer robots are cutting through the door. In retrospect, letting the inventor of Doom create an AGI may not have been the smartest move."
First define intelligence. Not artificial intelligence, real intelligence because unless we agree on that we can't tell whether you've achieved your goal in producing an artificial version. Not in some airy-fairy philosophical terms but in terms which has be independently confirmed and agreed upon. 2030? Good luck in achieving that first step by then. Otherwise you're simply putting whatever you've got in a box, calling it AI and claiming success.
I've just finished re-reading Feynman's appendix to the Challenger report. His last sentence is something that should be borne in mind by anyone making such claims:
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled."
Startup is named Keen
Does he by chance had given himself the role of Commander?
Specify the problem
"...thinks artificial general intelligence is doable by 2030"
Let's start by specifying what, exactly, AGI actually is and what we expect it to be able to achieve. Then maybe a sentence like the above could make sense
Those that look at LLMs like ChatGPT and decide AGI is 'only a few years away' are like a child that has seen a conjuror produce a coin from behind their ear and thinks they have found the solution to the national debt.